Full details are given in Jones et al. (2003) and Gillison et al. (2003). Soil Soil and vegetation samples were co-located for all sites in each region. Soils were sampled within the base transect and subjected
to routine laboratory analyses for a standard suite of parameters including texture, bulk density, pH, conductivity, C, N, P, S, exchangeable cations (Na, K, Ca, Mg), other mineral elements (Al, Mn, B, Zn, Cu, Fe) (Appendix S1, Tables S15–S18, Online Resources; see also Gillison 2000). Because most important soil information associated with plant and animal distribution is contained in the surface horizons, we report correlative analyses between soil data from 0 to 10 cm depth, and biota. Data analysis We examined whether simple measures of vegetation structure, and structural and functional trait diversity were meaningfully correlated with plant and animal species richness. The purpose was to identify PF299 mw straightforward and promising relationships that apply to diverse tropical communities, rather than single examples
where one biological feature predicts another. PFT data were analysed in two forms: https://www.selleckchem.com/products/ly3039478.html PFT counts per transect weighted by the number of species occurring in each PFT, and PFT counts recorded without reference to species (unique PFTs). In addition to whole PFTs, we disaggregated both PFT forms into their component elements (PFEs) to permit correlation of individual functional traits with individual species, species diversity and soil properties including carbon. Plants, birds, mammals and termites were assessed at individual species level and as assemblages. Sclareol To find easily
applicable indicators we focused on univariate linear relationships, as non-linear and multivariate relationships are more difficult to calibrate and apply, although we do not exclude the possibility that they occur (see Appendix S3, Online Resources). In a few cases we have reported quadratic univariate relationships that appear striking. Pearson product-moment analysis was used to generate a linear correlation matrix for all recorded variables for both regions separately and combined. Correlation was tabulated as the coefficient r and tested for significance via the Fisher-z transformation using Minitab 14.2 (Gillison 2005). Linear regression between pairs of variables was also carried out by the ordinary least squares method (1,307 regressions). In a few selected cases these are illustrated (Figs. 1, 2), with the equation of the fitted line and the adjusted coefficient of determination, RSq. In 160 cases of significant and 14 close-to-significant regression slopes, pairs of variables are tabulated with the t statistic (i.e. the slope of the line divided by its standard error) and its associated significance (Tables S21, S22, Online Resources). Fig. 1 Variations in correlative see more responses between animal taxonomic richness and plant-based indicators illustrated by birds and termites. The differences reflect regional ecosystem characteristics.